vene

"""
(C) August 2013, Mathieu Blondel
# License: BSD 3 clause

Custom group support by Vlad Niculae (vlad@vene.ro)

This is a Numba-based reimplementation of the block coordinate descent solver
(without line search) described in the paper:

    Block Coordinate Descent Algorithms for Large-scale Sparse Multiclass

vene

import re
from collections import OrderedDict
import numpy as np
from sklearn.base import BaseEstimator, TransformerMixin

class LexicalSetVectorizer(BaseEstimator, TransformerMixin):
    def __init__(self, word_sets=None, normalize=False, lower=False,
                 token_pattern=ur'(?u)\b\w\w+\b'):
        self.word_sets = word_sets
        self.normalize = normalize

vene

from __future__ import print_function

from sklearn.feature_extraction.text import CountVectorizer
from sklearn.grid_search import GridSearchCV
from sklearn.pipeline import make_pipeline
from sklearn.dummy import DummyClassifier
from sklearn.cross_validation import LeaveOneOut

docs = ["the cat lives in the hat", "the quick brown fox jumps over a dog",
        "a clockwork orange"]

vene

# Author: Vlad Niculae <vlad@vene.ro>
# Licence: BSD

from __future__ import division, print_function
import numpy as np
from sklearn.utils import check_random_state


class SquaredLoss(object):
    def loss(self, y, pred):

vene

#!/usr/bin/env perl

# Lemmatize CONLL-style (tabular) POS-tagged file using Treex
# Prerequisites: cpan -i -f Treex::Tool::EnglishMorpho::Lemmatizer
#   (I think the -f is needed because some tests are failing)
# Usage example:
#  $ echo "1\tgoes\t_\tVBZ\n" > example
#  $ <example ./lemmatize.pl
#  1    goes    go    VBZ
#

vene

The full tagger software package is licensed as GPL version 2.

src/ -- All original code we've written -- the files in src/ with one
exception below -- we license under the Apache License version 2.0. However,
we have several GPL'd dependencies that we include in this package, which,
as we understand it, force the full package to be GPL.

src/cmu/arktweetnlp/impl/OWLQN.java -- is licensed GPL, originally from the
Stanford POS Tagger version 2010-05-26.

vene

import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns

deaths = [596577, 142942, 73831, 41374, 39518, 21176, 7683, 6849]
money = [54.1, 7, 4.2, 257.85, 3.2, 147, 14, 22.9]
names = ["Heart disease", "COPS", "Diabetes", "Breast cancer",
         "Suicide", "Prostate cancer", "HIV/AIDS", "Motor neuron disease"]

sns.set_style("white")

vene

# Author: Vlad Niculae <vlad@vene.ro>
# License: 2-clause BSD

"""2D implementation of the robust Siegel Repeated Median slope estimator

This estimator tolerates corruption of up to 50% of the input points in either
the X or the Y dimension.

Vectorized implementation, and a naive implementation for sanity-check.
"""

vene

# -*- coding: utf-8 -*-
"""
Newton interpolation and numerical differentiation
Created on Sun Apr 10 01:22:46 2011

@author: vene
"""
from __future__ import division
from copy import copy

vene

import os

import numpy as np
import matplotlib.pylab as pl
from sklearn.svm import SVR
from sklearn.metrics import mean_squared_error

X = np.array([[13.],    # This is dataset no. 3 from Anscombe's quartet.
              [10.],    # I moved the outlier to the first position for
              [8.],     # prettier code. This toy dataset illustrates